init ingestion

ingestion/infra/clickhouse/000001_events.up.sql

@@ -0,0 +1,117 @@
+-- ---------------------------------------------------------------------------
+-- ClickHouse event store.
+--
+-- One wide table per event family (track / identify / page / group).
+-- All share the same key columns. Custom fields are flattened into the
+-- properties / traits Map.
+-- ---------------------------------------------------------------------------
+
+CREATE TABLE IF NOT EXISTS events_track
+(
+    workspace_id     String,
+    source_id        String,
+    message_id       String,
+    anonymous_id     String,
+    user_id          String,
+    event            String,
+    timestamp        DateTime64(3, 'UTC'),
+    sent_at          DateTime64(3, 'UTC'),
+    received_at      DateTime64(3, 'UTC'),
+    properties       Map(String, String),
+    context          Map(String, String),
+    ip               String,
+    user_agent       String,
+    library_name     String,
+    library_version  String
+)
+ENGINE = MergeTree
+PARTITION BY toYYYYMM(received_at)
+ORDER BY (workspace_id, source_id, received_at, message_id)
+TTL toDateTime(received_at) + INTERVAL 18 MONTH
+SETTINGS index_granularity = 8192;
+
+CREATE TABLE IF NOT EXISTS events_identify
+(
+    workspace_id     String,
+    source_id        String,
+    message_id       String,
+    anonymous_id     String,
+    user_id          String,
+    timestamp        DateTime64(3, 'UTC'),
+    sent_at          DateTime64(3, 'UTC'),
+    received_at      DateTime64(3, 'UTC'),
+    traits           Map(String, String),
+    context          Map(String, String),
+    ip               String,
+    user_agent       String
+)
+ENGINE = MergeTree
+PARTITION BY toYYYYMM(received_at)
+ORDER BY (workspace_id, source_id, received_at, message_id)
+TTL toDateTime(received_at) + INTERVAL 18 MONTH
+SETTINGS index_granularity = 8192;
+
+CREATE TABLE IF NOT EXISTS events_page
+(
+    workspace_id     String,
+    source_id        String,
+    message_id       String,
+    anonymous_id     String,
+    user_id          String,
+    name             String,
+    category         String,
+    timestamp        DateTime64(3, 'UTC'),
+    sent_at          DateTime64(3, 'UTC'),
+    received_at      DateTime64(3, 'UTC'),
+    properties       Map(String, String),
+    context          Map(String, String),
+    ip               String,
+    user_agent       String,
+    referrer         String,
+    path             String,
+    url              String
+)
+ENGINE = MergeTree
+PARTITION BY toYYYYMM(received_at)
+ORDER BY (workspace_id, source_id, received_at, message_id)
+TTL toDateTime(received_at) + INTERVAL 18 MONTH
+SETTINGS index_granularity = 8192;
+
+CREATE TABLE IF NOT EXISTS events_group
+(
+    workspace_id     String,
+    source_id        String,
+    message_id       String,
+    anonymous_id     String,
+    user_id          String,
+    group_id         String,
+    timestamp        DateTime64(3, 'UTC'),
+    sent_at          DateTime64(3, 'UTC'),
+    received_at      DateTime64(3, 'UTC'),
+    traits           Map(String, String),
+    context          Map(String, String),
+    ip               String,
+    user_agent       String
+)
+ENGINE = MergeTree
+PARTITION BY toYYYYMM(received_at)
+ORDER BY (workspace_id, source_id, received_at, message_id)
+TTL toDateTime(received_at) + INTERVAL 18 MONTH
+SETTINGS index_granularity = 8192;
+
+-- Dead-letter queue for events that failed validation / transformation.
+CREATE TABLE IF NOT EXISTS events_dlq
+(
+    workspace_id String,
+    source_id    String,
+    message_id   String,
+    received_at  DateTime64(3, 'UTC'),
+    reason       String,
+    field        String,
+    raw_payload  String
+)
+ENGINE = MergeTree
+PARTITION BY toYYYYMM(received_at)
+ORDER BY (workspace_id, received_at)
+TTL toDateTime(received_at) + INTERVAL 30 DAY
+SETTINGS index_granularity = 8192;